On Jan 8, 2010, at 10:41 AM, Brice Goglin wrote: > > If we want to whack the .tex files, we certainly can (and just include the > > HTML and PDFs). I don't think I have a strong feeling either way. > > Are these tex files shipped because people might want to rebuild the PDF > ? I am ok if we don't ship them.
Yes, and I agree. I'll work on whacking them in the build process. > But we still have to fix the HTML and manpage long filenames. We could: > * use shorter type names in the device attributes, but it'll be hard > anyway since we it's waaaaaay too long. doxygen didn't actually go the > lastest level of nested struct/unions: we have union hwloc_obj_attr_u -> > struct hwloc_bridge_attr_u -> union hwloc_bridge_downstream_attr_u -> > struct hwloc_bridge_downstream_pci_attr_u > * stop doing nested declarations of all these attribute union/structs so > that doxygen generates foo and bar instead of foo and foo_bar. but we'll > end up having many struct/union declarations and I think it would be > hard for people to read them while traversing the unique nested > declaration is easy. > > I am looking for a way to tell doxygen to output foo and bar even when > bar is declared inside foo but I didn't find anything yet. I don't think we should let doxygen dictate the code that we write. I see 3 obvious solutions: 1. make doxy do what we want 2. post-process the doxy output to have more friendly filenames 3. use some other tool -- Jeff Squyres jsquy...@cisco.com